Researchers Crack KeeLoq Code for Car Keys: Slashdot had a small post about this yesterday but I've spoken with one of the researchers about it so thought I'd add a little more detail.

A group of Israeli and Belgian researchers found a vulnerability in the algorithm that is used to secure anti-theft digital key systems in numerous vehicles, including those made by such companies as Honda, Ford, General Motors, Mercedes Benz and Jaguar. With that information they were able to devise an attack to crack the code of anti-theft keys.

With just an hour of remote access to the digital key of one car made by a manufacturer, the researchers say they are able not only to crack the unique code for that specific key but can also determine the key initialization process used to code the digital keys for all of the cars made by that manufacturer. From there, it's pretty simple for them to crack the unique code of another car made by that company.

'There is one master key from which is derived the key for each car a company makes,' says Orr Dunkelman, a researcher from the University of Leuven in Belgium who worked on the project with four colleagues.

The cipher the researchers examined, known as KeeLoq, is used to lock and unlock many vehicles that employ remote-control key-less systems and digital key systems -- key fobs and keys that are embedded with a chip that transmits a unique digital code. The devices not only lock and unlock car doors but can also mobilize or immobilize a vehicle and operate its alarm system.

The KeeLoq technology, which is licensed by Microchip Technology to car makers and other entities, has long been considered to be pretty secure. Each KeeLoq key or key fob uses a unique value, out of billions and billions of possibilities, to unlock a car.

But after proprietary information about KeeLoq was leaked to a Russian hacking web site (pdf) last year, the five researchers, from the University of Leuven as well as the Hebrew University and the Technion in Israel, began examining the system for vulnerabilities. Within three to five days Dunkelman says they developed their first basic attack, then spent months refining their technique.

The attack involves probing a digital key wirelessly by sending 65,000 challenge/response queries to it. Once the researchers collect 65,000 responses – which takes about an hour – they use software they designed to decipher that key's unique code. The deciphering currently takes about a day using a dedicated computer. But once they've cracked one key, they know 36 bits of the 64 bits they need to know. Those 36 bits are identical for every car model a manufacturer makes (different car models will vary only slightly).

This doesn't mean Dunkelman can just walk onto a parking lot and open any car that's the same model of the one he cracked. He still needs to crack the unique key used to open the other cars. But because he already knows the 36 bits that are common to all of the keys for one model of car, it takes only a few seconds to crack those other keys. He can do this by reading the keys wirelessly -- for example, while sitting next to a patron at a restaurant or standing near a car when an owner opens it and sniffing the communication between the digital key and the car. Once he has a key's unique code, he can encode it to a chip in a remote device (which he can do in a couple of seconds in the field) and use it to open and steal the car.

Dunkelman says an ideal scenario might be for a valet attendant to sniff the keys of cars in his charge to obtain the 36 bits for several different makes and models.

'If you go to a restaurant and leave your car with valet parking, the parking guy can probe your key while you're eating and enjoying your steak and by the time you leave he has all the information he needs,' says Dunkelman.

Last week Dunkelman and his colleagues contacted Microchip Technology to report their findings. They also discussed the findings at the Crypto conference. But they say they won't release their research paper publicly until after they've heard from Microchip.

A spokeswoman for Microchip declined to comment on the researcher's findings.

(Read Original Article - Via Threat Level.)